University of Dundee

"Mathematical Models for Bioimage Analysis"

Event Date: 
Friday, February 7, 2020 - 11:00
Event Location: 
School of Life Sciences, MSI Small Lecture Theatre
Host: 
Dr. Gabriele Schweikert
Professor Jason Swedlow FRSE
Event Speaker: 
Dr Virginie Uhlmann
Institution: 
European Bionformatics Institute (EMBL-EBI), Cambridge
Event Type: 
Seminar
Share

Abstract:

Images and video sequences are crucial sources of information in experimental biology, and bioimage analysis has grown into an independent field of research. Recent advances in machine learning provide efficient tools which, given a set of examples, learn to identify parts of image data that hold relevant information. By construction, images are however bound to a grid representation, which is ill-suited for characterizing and modeling biological objects. While the process of image acquisition acts as an analog-to-digital converter, projecting real objects onto the space of image data, parametric curve and surfaces models can be used to revert this effect and characterize biological object from images in an “infinite resolution” way. My group focuses on the development of such models to offer a common, unified mathematical representation of morphology, regardless of the imaging modality. Ultimately, it allows studying phenotypic features, as well as changes and differences thereof, in a purely analytical way. In this talk, I will describe the general principles behind the construction of parametric curve and surface models and illustrate their use in various bioimage analysis problems.